|Publication number||US7519534 B2|
|Application number||US 10/699,543|
|Publication date||Apr 14, 2009|
|Filing date||Oct 30, 2003|
|Priority date||Oct 31, 2002|
|Also published as||US20040193426|
|Publication number||10699543, 699543, US 7519534 B2, US 7519534B2, US-B2-7519534, US7519534 B2, US7519534B2|
|Inventors||Scott Lynn Maddux, Adam Jordan, Jody Williams, Harry W. Printz|
|Original Assignee||Agiletv Corporation|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (78), Referenced by (22), Classifications (18), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application claims priority and incorporates by reference the Provisional Application No. 60/422,561, filed Oct. 31, 2002 and the Provisional Application No. 60/483,268, filed Jun. 26, 2003.
1. Technical Field
The invention relates to user access to content. More particularly, the invention relates to speech controlled access to content on a presentation medium.
2. Description of the Prior Art
In recent years, the number of channel and program choices for cable and digital broadcast satellite subscribers has grown to the point where the television viewer is faced with a vast array of programming options. The breadth of service providers and depth and variety of content has made it more difficult for the television viewer to find the programs they want to watch. In the face of this challenge, the television viewing experience remains anchored in passive simplicity.
It would be advantageous to provide television viewers with an intuitive and easy-to-use way to find the programs they want and to control their television viewing experience.
It would also be advantageous to provide a speech control interface, in combination with a variety of search functions, such that television viewers have an intuitive and easy-to-use way to find the programs and information they want.
It would also be advantageous to provide the use of personalization as a way to filter and deliver relevant content and services to users based on individual preferences.
It would also be advantageous to provide a hierarchical user interface for speech-controlled, interactive applications, that gives television viewers an intuitive and easy-to-use way to find the programs they want and to control their television viewing experience.
Another problem exists in existing digital cable systems, where the channels that are available to watch in full screen TV and the program guide (the “channel lineup”) are fixed and non-changing, assuming that the channel lineup itself does not change. In current systems, when the channel up and down buttons are pressed, the user navigates through all channels of the entire channel lineup.
It would be further advantageous for the channel lineup to be dynamic and changeable, i.e. the list of available channels can be modified, in real time, to suit the preferences and usage of a particular subscriber.
There is also substantial prior art in the design of speech recognition systems, and in systems that identify the speaker, group, language, or emotional state associated with a given speech sample. There is also substantial prior art in selecting an advertising medium or vehicle to target a desired segment of the population.
It would be advantageous to provide a method and apparatus that addresses the use of speaker, group, language, or emotional state identification systems to target advertising, to users who are enrolled in a speaker ID system, by virtue of having previously supplied a voice sample to the system, or to users who are not so enrolled.
One embodiment of the invention provides television viewers with an intuitive and easy-to-use way to find the programs they want and to control their television viewing experience. In a further embodiment, a speech control interface, in combination with a variety of search functions, is provided such that television viewers have an intuitive and easy-to-use way to find the programs and information they want. The invention also provides the use of personalization as a way to filter and deliver relevant content and services to users based upon individual preferences. Another aspect of the invention provides a hierarchical user interface for speech-controlled, interactive applications, that gives television viewers an intuitive and easy-to-use way to find the programs they want and to control their television viewing experience. The invention also provides a method and apparatus where the channel lineup in a television service is dynamic and changeable, i.e. the list of available channels can be modified, in real time, to suit the preferences and usage of a particular subscriber. Further, the invention provides a method and apparatus that addresses the use of speaker, group, language, or emotional state identification systems to target advertising, to users who are or are not enrolled, by virtue of having previously provided a voice sample to the system.
One embodiment of the invention provides television viewers with an intuitive and easy-to-use way to find the programs they want and to control their television viewing experience. In a further embodiment, a speech control interface, in combination with a variety of search functions, is provided such that television viewers have an intuitive and easy-to-use way to find the programs and information they want. The invention also provides the use of personalization as a way to filter and deliver relevant content and services to users based upon individual preferences. Another aspect of the invention provides a hierarchical user interface for speech-controlled, interactive applications, that gives television viewers an intuitive and easy-to-use way to find the programs they want and to control their television viewing experience. The invention also provides a method and apparatus where the channel lineup in a television service is dynamic and changeable, i.e. the list of available channels can be modified, in real time, to suit the preferences and usage of a particular subscriber. Further, the invention provides a method and apparatus that addresses the use of speaker, group, language, or emotional state identification systems to target advertising, to users who are or are not enrolled in a speaker ID system by virtue of having previously supplied a voice sample to the system (see below).
The foregoing embodiments of the invention are discussed in detail below:
This embodiment of the invention provides a method for constructing grammars for speech-controlled, interactive applications that provides television viewers with an intuitive and easy-to-use way to find the programs they want and to control their television viewing experience. An interactive application is one in which more than one party is an active participant. For this embodiment, one of the parties is human and the other is a software application. There are various publications describing how to construct a grammar, available from the VoiceXML Forum and other sources.
Grammars place constraints on the sequence of words allowed in utterances, which improves recognition performance. The number of words and phrases that are used to describe, for example, television-related content is so large that they must be carefully organized to give users high recognition accuracy when making verbal requests to control their television sets or to search for television-related information. This embodiment of the invention organizes the grammars in a way that provides high recognition accuracy for utterances from a large and continually expanding set of words and phrases. It does this without requiring users to provide samples of their voices ahead of time to train the system.
The information space is organized by information type, such as actor or TV program for television-related content. The following discussion describes how these information types can be used to create a scalable, flexible basis for developing interactive speech-controlled applications. The information architecture allows information to be linked or organized into a hierarchy, and gives the application access to the path that the user chose to navigate through the information. The information in the grammars can also be used to carry out requests initiated by buttons on a remote control or other such device.
Each speech-controlled application uses a set of commands initiated by either speech or input from a more traditional device, such as a remote control or pointing device. Speech commands are specified in a set of grammars. For the purposes of this discussion, grammars contain a set of items. Each item contains a word or phrase that can be spoken, an associated set of attributes that provide more information about the item, and optionally, the name of a grammar that must be imported to complete the utterance. A grammar also has an associated set of attributes that provide more information about the grammar as a whole.
Grammars that are largely composed of utterances that are used to control an application are referred to as command grammars in the rest of this disclosure, and grammars that contain multiple alternatives for a single piece of information, such as a set of actor names, are referred to as information-type specific grammars.
An information-type specific grammar may be linked to another grammar via a unique utterance, referred to as a keyword, that distinguishes this information type from any other. These grammars have an attribute that specifies this keyword, and the items in the grammar contain an attribute that identifies the item's information type.
Some speech commands may be chained together to provide one-step access to information or application functionality. The Chained Command attribute indicates that the utterance in an item is a part of a chained command. The Chained Command attribute may appear in both command and information-type specific grammars.
The keywords used to chain grammars together should be acoustically distinct. That is, the keywords should not sound similar to other keywords or other utterances in the grammars. The more distinct they are, the better recognition accuracy is.
Note that user input may also be received from a pointing device or remote control.
Grammars constructed using this information architecture enable easy access to information hierarchies, allow multiple linked navigation paths, and facilitate the development of more flexible applications.
Grammars can be linked into a hierarchy as follows:
An utterance that links two grammars is defined in the grammar in two parts. The first part is a keyword, for information-type specific grammars, or a command for command grammars. The contents of the grammar-to-be-linked are imported to include all of the alternatives for the second part of the utterance. As long as the keyword or command is not one of the items in the higher-level grammar, this ensures that linking the two grammars does not significantly degrade recognition accuracy.
Application writers can chain grammars together to give users one-step access to application functionality. This is shown in
“Set”, Chained Command, “Options”, Chained Command, “Favorite”, Chained Command, “Channel 2” , Channel Type. The values returned in the second case are: “Set” , Chained Command, “Options” , Chained Command, “Favorite” , Chained Command, “Category”Chained Command, “Sports”, Category Type.
Linked Navigation Paths
Grammars can link information-type specific information either explicitly or implicitly.
An application can predetermine the navigation path and link in all of the appropriate grammars. This is shown in
The application may also rely on the type attributes returned by the recognizer to determine which options to make available to the user next, as in
The grammars shown in
The grammar in
This approach can also be used to support the “more like this” command. The recognizer returns “more like this,” and the application uses the currently selected item type to decide where to look for the information that the user wants.
Flexible Application Structure
This information architecture of the presently preferred embodiment can be used, for example, in two different ways to extend existing applications with minimal development effort.
First, some of the user interface for an application can be built dynamically from the data in the grammars. The grammars contain the structure of the speech commands for the application, and indicate when commands are chained. Grammars may be parsed to obtain this structure and the structure may be used to auto-generate some of the user interface for the application.
Second, it is easy to extend the content accessible via the application. A new link can easily be added to any of the chained commands.
This invention has been reduced to practice via implementation in C-code, XML grammar files, compilation, and the generation of executable images for the AgileTV engine (AgileTV, Menlo Park, Calif.) to implement a very simple program name search. The program name keyword was added to a command grammar in an existing application, and the name of every program on the Palo Alto, Calif. cable television system in a two-week period was added to a program-specific grammar. The program-specific grammar contained approximately 7000 utterances, and recognition accuracy for the application as a whole was not significantly affected.
The method and system described above also has application in support of the following services:
This embodiment of the invention provides a speech control interface in combination with a variety of search functions which provide television viewers with an intuitive and easy-to-use way to find the programs and information they want. Users typically search and/or discover new programs in one of two ways:
The specific method described in this document is designed to provide a simple user experience and maintain high recognition accuracy for the speech control interface.
Exploratory search is a method by which the user can find additional data with similar attributes to an already known piece of data. Within the space of television-related information, exploratory search allows users to find other programs related in some manner to a currently-selected program, such as finding more movies starring the same actor who is in a movie that the user is currently watching. This type of searching affords more free-form discovery, in that specific information types and items may not be known prior to the search. Exploratory searches are controlled with hierarchical or chained speech commands, or on-screen buttons. Results from exploratory searches are displayed in a list on the television screen.
Exploratory searches may be controlled explicitly or implicitly.
In an explicit exploratory search, the user specifies a particular information type by which the search should be performed. For example, while watching a movie on television, the user may ask for other movies by the same director using the appropriate speech command, e.g. “Ron Howard.” A list of movies by that particular director is then displayed in an on-screen list. Alternatively, while in the program guide, the user may ask to see other programs starring the same actor as is in the selected program.
Information types for explicit searches of television programs include, but are not limited to: Actor, Director, Writer, Creator, Episode, Date, Popularity (Nielsen), Quality (Editorial/Star Ratings).
In an implicit exploratory search, the user asks for more data related to the currently selected data. The specific information types are determined programmatically by the search application. For example, if program information for a movie is displayed, and the user performs an implicit exploratory search using the appropriate command, e.g. “more like this,” the search application finds other movies with similar information items, such as the same genre, director, or actors.
Another example of an exploratory search is a search that is based on recommendations. This search method is used when the television viewer chooses to rely on a trusted source for a recommendation, e.g. Ebert & Roeper's two-thumbs-up selections or The American Film Institute's Top 100 movies.
Targeted Search Criteria
Targeted search is a method by which the user can find specific data, such as program names and movie titles, according to particular criteria. Within the space of television-related information, targeted search allows users to find specific programs according to attributes of those programs. Using a Speech-Optimized Information Architecture and Hierarchical Speech UI (as described herein), the user initiates a targeted search by specifying an information type and one or more items of that type. The targeted search then returns programs that match the specified item or items in a list that is displayed on the television screen. The control of targeted search may be accomplished through both one-step, chained speech commands, as well as by progressive multi-step speech commands and on-screen buttons.
A targeted search may be performed on any data that belongs to one or more information types. Within the television space, information types include, but are not limited to: program name, genre, actor, director, writer, episode, date, popularity (Nielsen), quality (editorial, star rating), and subject matter rating (MPAA ratings and TV Ratings).
Using the Hierarchical Speech UI, targeted search is modeless, ubiquitous, and always accessible. It can be performed in any mode and on any screen. For example, if the user wants to see if any movies starring Mel Gibson are airing in the evening, the targeted search may be initiated while watching TV or in any screen in the program guide using the appropriate command, e.g. “Find Actor Mel Gibson.” Additionally, if the user wishes, targeted search may also be controlled through progressive search modes and screens that do not require the memorization of any commands or keywords.
The method and system described above has application in support of, for example, the following services:
A user utters commands via a voice link remote 81. These commands are received at a voice link receiver 82, which in the presently preferred embodiment, is coupled to a cable television set top box 83. The commands are processed and forwarded via a hybrid fiber-coax cable plant 84 to a return path receiver 85, such as that provided by AgileTV of Menlo Park, Calif. A processor, such as a Linux-based processor 86, supports speech recognition and processes the user's utterances into recognizable commands. A path back to the user for a system response is provided by a QAM modulator 87, such as manufactured by AgileTV of Menlo Park, Calif. The system, responsive to user commands thus received, accesses various forms of content for presentation to the user, such as a video-on-demand server 90, an interactive program guide server 89, or a walled garden server 88 which provides the user with various Internet applications.
Personalization Process, Personalized Program Lineup, Recommendations, and Interactive Television Services
This embodiment of the invention provides the use of personalization as a way to filter and deliver relevant content and services based on individual preferences. In turn, this limits the number of options faced by a user at any given point and simplifies the user experience. Voice recognition and the speech interface provide an intuitive way to navigate the interactive television interface and manage the thousands of options from which to choose.
The following steps are performed to provide a personalized experience to the individual television viewer:
Step 1. Preference Profiling
To create a preference profile for an individual television viewer a variety of methods may be used, including the following:
Step 2. Identification
Identification of an individual viewer may be accomplished in one of two ways:
Step 3. Preference Matching
Preference matching takes place once the individual user has been identified or grouped and a preference profile is matched to the user and activated. This is a simple database matching procedure 96.
Step 4. Presentation
Presentation of personalized information is the final step in which programming content and/or interactive television services are displayed for the individual user via the graphic user interface and controlled via speech recognition. Personalized programming content and information services may take the following forms:
The method and system described above also has application in support of the following services:
This embodiment of the invention provides a hierarchical user interface for speech-controlled, interactive applications that provides television viewers with an intuitive and easy-to-use way to find the programs they want, and to control their television viewing experience. An interactive application is one in which more than one party is an active participant. For this disclosure, one of the parties is human and the other is a software application.
The hierarchical user interface provides modeless, ubiquitous, one-step access to information content and other application functionality via speech commands. It also discloses a graphical user interface that gives progressive disclosure to the same information content via input from more traditional sources such as a pointing device or remote control. Speech commands can also be used one at a time to follow the progressive disclosure path.
Each speech-controlled application uses a set of commands initiated by either speech or input from a more traditional device such as a remote control or pointing device. Speech commands are specified in a set of grammars. There are various publications describing how to construct a grammar, available from the VoiceXML Forum and other sources. For the purposes of this discussion, grammars contain a set of items. Each item contains a word or phrase that can be spoken, an associated set of attributes that provide more information about the item, and optionally, the name of a grammar that must be imported to complete the utterance. A grammar also has an associated set of attributes that provide more information about the grammar as a whole.
This embodiment of the invention allows information to be linked into a speech-accessible hierarchy. The hierarchy is implemented as a grammar. Any information or application functionality that should be ubiquitous, that is, available from any of the application screens or modes via speech commands, is described in grammars that are linked into the hierarchy.
The application then includes the hierarchy in the set of grammars activated by the Recognizer for any speech command. All of the speech commands contained in the hierarchy are then available from everywhere within the application. The reason for making commands ubiquitous is that it is frequently awkward to navigate to a user interface element with which the user wants to interact. The example in
In a hierarchical user interface, application settings can be implemented as chained commands linked into a hierarchy, e.g. from anywhere in the application, “Set Options Favorite Channel 2” to add the favorite channel in a single step.
The graphical user interface can be constructed to mirror the structure of the hierarchy so that there is a visual element representing each command in the chain. To reflect the hierarchy, the visual elements are usually disclosed one at a time. When the user engages one of these visual elements, the visual element representing the next level in the hierarchy appears. The menu example in
An example user interface for the same step-by-step functionality in a speech-enabled television context is shown in
The grammars can also be constructed to allow the user to speak each of the commands in the chain as an individual command. At each link in the chain, both the top-level hierarchical grammar and a grammar for the sub-hierarchy that begins with the keyword for this level must be in the set of grammars activated by the Recognizer.
The visual elements representing the commands in a chain can take many different forms. They can be buttons, items in menus, or text in speech feedback given to the user. Each visual element should include the utterance used in the grammar so that when part of a chained command has been disclosed, the user can pick up where they left off in the chain. For example, if the user says “Find Sports” and then sees a visual element representing “Team,” he can then say “Team Bengals.” If these commands are spoken while the user is watching television, the feedback appears in an information bar, for example as shown in
The hierarchical user interface also makes it easier for users to learn which speech commands are accessible in which application contexts.
The method and system described above also has application in support of the following services:
This embodiment of the invention provides for the channel lineup to be dynamic and changeable. That is, the list of available channels can be modified, in real time, to suit the preferences and usage of a particular subscriber.
A dynamic channel lineup allows the user to change the set of channels that are available to watch and to get program listings for. Channels can easily be removed and added to the channel lineup at any time using speech commands. The control of the dynamic channel lineup is immediate and modeless. That is, there is no need to change to a different mode or go to a particular setup screen to change the lineup. Control of a dynamic channel lineup could also be achieved with buttons.
Additionally, the control of the channel lineup allows both temporary and more permanent changes to the lineup. That is, some changes to the lineup can be engaged for a short period of time, such as one period of viewing, while other changes to the lineup remain in effect until they are disabled.
When a dynamic channel lineup is in effect and the user is watching TV, the channel navigation commands only show the channels within the dynamic lineup. When the user is in the program guide application, only listings for the channels in the dynamic lineup are shown.
There are five methods of controlling a dynamic channel lineup:
By using the personal profiling and speaker identification technologies, individual users can create and use a personal channel lineup. This personal channel lineup is activated when the user issues the appropriate command, e.g. “show my channels.” If the user is enrolled in speaker identification and/or has a personal profile, the channel lineup is specific to that particular user. If the user is not enrolled in speaker identification or does not have a personal profile, a default, household channel lineup is activated.
Once the personal channel lineup is activated, only the channels that are included in the lineup are available. A user may add or remove channels from the personal lineup by issuing the appropriate commands, e.g. “add HBO”, “remove QVC.”
Particular programs can also be added to the personal lineup, in which case the associated channel is temporarily added to the personal lineup for the duration of the program.
Category Channel Lineups
A category channel lineup is created using particular categories of programs. The user activates the category channel lineup by issuing the appropriate command and a category name, e.g. “show movie programs.” Once the category lineup is in effect, only channels that are currently airing programs of the specified category are available.
Different types of program categories include:
Of particular interest is a channel lineup comprised only of child-appropriate content. This channel lineup, called “Kids only,” consists of programs from child-related genres with an appropriate TV rating. The “Kids only” lineup acts as a default parental control. Parents need only to say the command, e.g. “Show Kids only,” to activate the Kids only lineup.
Category channel lineups may also consist of combinations of categories. Those categories may be related, such as family and comedy, or unrelated, such as baseball and fishing.
Channel. Program, Rating, and Category Blocks
There are four different methods of removing (“blocking”) programs and channels from a dynamic lineup:
Each method, described below, either permanently or temporarily makes a channel unavailable while watching TV or viewing listings in the program guide application.
A particular channel may be globally removed from a dynamic channel lineup. When the Block command and the channel name, e.g. “block HBO,” is given, the specified channel is no longer available while watching TV or viewing listings in the program guide. Any channel may be blocked at any time, regardless of whether or not the channel is currently tuned to or selected with the program guide.
Channels may be unblocked by issuing the Unblock command and the channel name, e.g. “unblock HBO.”
Programs can be blocked using a similar mechanism as described above. When a program is blocked, e.g. “block Sex in the City,” the channel that it is airing on is unavailable for the duration of the program. As with channels, programs may be unblocked with the appropriate command (“unblock Sex in the City”).
Programs can also be blocked by a specified category, such as “adult” or “shopping,” or by a specified MPAA or TV rating, such as “PG-13” or “TV-R.” Category and rating blocks temporarily remove channels from the dynamic lineup when programs that match the category or rating are airing or are within the current program guide time slot. The channels remain unavailable for the duration of the program.
Blocks can either be persistent across other dynamic channel lineups, such as personal lineups or category lineups, or can be set independently within a personal channel lineup.
Atrophying Unwatched Channels
If channels are never watched for a particular time interval, such as four weeks, they can automatically be removed from the dynamic channel lineup. This control can also be forced by the user with the appropriate command, e.g. “remove unwatched channels.”
Channel Lineup Reset
The dynamic channel lineup can also be reset, causing all channels to become available. This global reset is activated by issuing the appropriate command (“reset all channels”).
The channel lineup reset may also be selectively applied to a personal channel lineup with a different command (“reset my channels”).
This embodiment provides a method and apparatus that addresses the use of speaker, group, language, or emotional state identification systems to target advertising, to either enrolled or unenrolled users of a content presentation system. The invention pertains to improving the effectiveness of electronic advertising, typically though not exclusively when presented to the consumer via television, or via a computer connected to the Internet. Specifically, the invention applies to the case where this presentation medium, i.e. the television, Internet-connected computer, etc., is controlled by spoken command.
Control by spoken command may be achieved through use of a speech recognition system, which processes an electronic representation of an audio signal, uttered by the user, into corresponding text, or into an associated signal that controls the presentation medium, i.e. tuning the television to a selected program, or causing the Internet-connected computer to display a selected Web page. It is key that the same information that is present in the audio signal, which can be processed to determine the spoken words or command, can also be used to determine the identity of the speaker. This latter determination is made by a speaker identification system.
The general context of this invention is shown in
Typically, though as detailed below not exclusively, a person wishing to use a speaker identification system.(also known as a speaker ID system) first provides a sample of his speech, and associates this speech sample with his personal identity, i.e. a name, credit card number, or other personal information. This process is known as enrollment. Thereafter, when the enrollee speaks to the speech processing apparatus, both the text corresponding to the utterance, and the identity of the speaker, is determined by the combined operation of the speech recognition and speaker identification systems. The mathematical representation of the enrollment data, which may be used to identify the speaker, is called a speaker identification model.
The general structure of the pairing of a speaker identification model with personal identity is illustrated in
The appeal of a speaker identification system is that it provides a secure and easy means for a user to authorize and control access to the presentation medium in question (television, computer, etc). Thus the user is likely to wish to enroll.
It is important to note that, after enrollment, a user's command utterance provides both command and control information, for example, selection of the program that the television is to tune to, and also the identity of the speaker. No special action by the user is required after the enrollment stage. The identity of the speaker is determined and reconfirmed with each command utterance. This identification is performed for each utterance, in a manner that is transparent to the user.
Users who have participated in the enrollment process are referred to as enrolled speakers. All others are referred to as unenrolled speakers.
One problem addressed by this invention is improving the effectiveness of electronic advertising. Advertising in general is a mass medium, which is to say that its message is widely disseminated throughout the consumer population. As such, it is inefficient, because the advertiser pays to blanket the entire audience with a commercial or political message, whereas only a small fraction of the audience may have the means or desire to respond to the message. For instance, a political message impelling voters to elect this or that candidate is wasted on viewers of the advertisement who are too young to vote.
For this reason, advertisers typically aim to promote their products or ideas through a means that is watched or used by a population that most closely matches the intended recipient of the advertising message. Thus, children's toys are advertised on television programs that are watched by primarily by children, beer commercials are shown on sports programs, and so on.
However, this selection mechanism, which is based on the general demographic appeal of the content being viewed, is less refined than might be desired. This embodiment of the invention provides a method for achieving far higher accuracy and precision in targeting of advertising.
The exemplary embodiment of the invention consists of targeting advertising content to individual users, as determined by the speaker identification system. In the context associated with this invention, the speech recognition system is used to select the content that the user wishes to experience, i.e. a television program, an audio stream of a particular genre, and so on. The pauses or interruptions present in this program or stream may then be filled with advertising messages that are targeted to the characteristics of the individual user who made the program selection.
This system permits the targeting of advertising by age, gender, income level, residential address, and other information that is highly pertinent, from the advertiser's point of view. This is because the speaker ID enrollment process typically, though not necessarily, links the user's audio sample with pertinent demographic or financial information, such as age, gender, home address, credit card number, and so on. Thus, for example, it would be possible for advertisers to ensure that their message is viewed by all households, and only those households, where an adult with a credit card credit debt limit of $10,000 or more is currently watching the television.
Discussion, Refinement and Extensions
The following are alternative embodiments of the invention, as presented above.
Targeting of Advertising without Speaker Identification Enrollment.
The system as described above assumes that an individual user has performed speaker ID enrollment, as defined in the opening paragraphs of this disclosure. However, it is possible to apply the same general idea in the absence of enrollment. This is because it is frequently possible to make an approximate determination of a speaker's age and gender, even absent an enrollment process. This mode of usage does not provide the individual targeting described above, but it still provides some of the audience focus that an advertiser seeks.
Targeting in the absence of enrollment is achieved by collecting samples of speech of adult males, adult females, adolescent males, adolescent females, and so on from the general population, and then building a general adult male speaker identification model, a general adult female speaker identification model, and so on.
As indicated in
Targeting of Advertising Based upon Emotional State.
By analyzing a spoken utterance for pacing, stress patterns, spectral distribution of acoustic energy and other elements, it is possible to make an approximate determination of a user's emotional state. Again, this can be valuable information to an advertiser. An individual who is agitated could be shown advertisements for stress-reducing products (herbal teas, spa visits, vacations, and so on). Those in an emotional state that is associated with an increased propensity to make major purchases could be shown advertisements for expensive or luxury items.
Targeting of Advertising Based upon Language.
In a further extension of this idea, it is possible to determine, again by analysis of the acoustic signal, the language in which a given utterance is spoken (English, Spanish, Hindi, etc). When the language of the speaker has been identified, an advertisement in that same language, and/or with content targeting speakers of that language, may then be presented to the user, at an appropriate interval. Thus for instance, Spanish speakers may be presented a Spanish-language version of an automobile commercial, or Hindi speakers may be presented advertisements for reduced-price air fares to and from the Indian subcontinent.
Combination with other Modes of Demographic Targeting.
It is well-known that the audience of particular television programs, or the users of particular Internet sites, may constitute a population segment that is sought by a particular advertiser. Indeed, this is the basis for many current techniques for demographic modeling. In a further extension of these ideas, by tracking the television program or website selections of an individual user, where that user has enrolled in the speaker identification system, and it is permissible by statute or explicit granting of permission by the user for this information to be recorded, it is possible to create an even more detailed demographic portrait of the user. Thus for instance, it would be possible for an advertiser to target all 35-to-45-year-old males who are frequent viewers of the television programs The West Wing and E.R.
Use of Blind (Unenrolled) Individual Speaker Identification.
As observed above, in general a speaker identification system requires that a user participate in an enrollment process, to capture a sample of the user's speech, and associate it with personal identity information. However, given sufficient acoustic data, captured through ordinary operation of the system by spoken command, as opposed to explicit enrollment, it is possible to distinguish among the individual users of the system, and build speaker identification models for them. These models do not have explicit personal identity information associated with them, by which is meant name, credit card number, etc. However, they nevertheless function to distinguish one frequent user of the system from another. Note that this differs from merely identifying the group to which a speaker belongs, as discussed above in connection with Targeting of Advertising without Speaker Identification Enrollment.
With this identification in hand, and to the extent that it is permissible by statute or explicit granting of permission by the user for this information to be recorded, a television or Internet selection history for such individual users may be created. This may be supplemented with a plausible guess of the user's gender, approximate age, and so on, by the methods previously described. In this way, a detailed demographic portrait of an individual user may be created, even absent explicit enrollment in the speaker identification system.
General Speech Characteristics.
To the extent that any other desirable identifying characteristic, from the standpoint of efficacy of advertising, may be extracted from a speech signal, this system may be extended to incorporate use of that characteristic, for the purpose of targeting advertising.
Although the invention is described herein with reference to the preferred embodiment, one skilled in the art will readily appreciate that other applications may be substituted for those set forth herein without departing from the spirit and scope of the present invention. Accordingly, the invention should only be limited by the Claims included below.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5752232 *||Nov 14, 1994||May 12, 1998||Lucent Technologies Inc.||Voice activated device and method for providing access to remotely retrieved data|
|US5774859 *||Jan 3, 1995||Jun 30, 1998||Scientific-Atlanta, Inc.||Information system having a speech interface|
|US6130726 *||May 15, 1998||Oct 10, 2000||Evolve Products, Inc.||Program guide on a remote control display|
|US6298324 *||Nov 12, 1998||Oct 2, 2001||Microsoft Corporation||Speech recognition system with changing grammars and grammar help command|
|US6301560 *||Jan 5, 1998||Oct 9, 2001||Microsoft Corporation||Discrete speech recognition system with ballooning active grammar|
|US6320947||Sep 14, 1999||Nov 20, 2001||Satyam Enterprise Solutions Limited||Telephony platform and method for providing enhanced communication services|
|US6374177||Sep 20, 2000||Apr 16, 2002||Motorola, Inc.||Method and apparatus for providing navigational services in a wireless communication device|
|US6374226 *||Aug 6, 1999||Apr 16, 2002||Sun Microsystems, Inc.||System and method for interfacing speech recognition grammars to individual components of a computer program|
|US6381316||May 9, 2001||Apr 30, 2002||Unpaid Systems, Ltd.||Enhanced communication platform and related communication method using the platform|
|US6408272 *||Apr 12, 1999||Jun 18, 2002||General Magic, Inc.||Distributed voice user interface|
|US6415257 *||Aug 26, 1999||Jul 2, 2002||Matsushita Electric Industrial Co., Ltd.||System for identifying and adapting a TV-user profile by means of speech technology|
|US6658414||Mar 6, 2001||Dec 2, 2003||Topic Radio, Inc.||Methods, systems, and computer program products for generating and providing access to end-user-definable voice portals|
|US6711543||May 30, 2002||Mar 23, 2004||Cameronsound, Inc.||Language independent and voice operated information management system|
|US6714632||Apr 3, 2002||Mar 30, 2004||Upaid Systems, Ltd.||Enhanced communication platform and related communication method using the platform|
|US6721633||Mar 1, 2002||Apr 13, 2004||Robert Bosch Gmbh||Method and device for interfacing a driver information system using a voice portal server|
|US6725022||Sep 20, 2000||Apr 20, 2004||Motorola, Inc.||Method and apparatus for enabling the selection of content on a wireless communication device|
|US6728531||Sep 20, 2000||Apr 27, 2004||Motorola, Inc.||Method and apparatus for remotely configuring a wireless communication device|
|US6799201||Sep 19, 2000||Sep 28, 2004||Motorola, Inc.||Remotely configurable multimedia entertainment and information system for vehicles|
|US6892083||Aug 30, 2002||May 10, 2005||Vocera Communications Inc.||Voice-controlled wireless communications system and method|
|US6901366 *||Aug 26, 1999||May 31, 2005||Matsushita Electric Industrial Co., Ltd.||System and method for assessing TV-related information over the internet|
|US6975993 *||May 2, 2000||Dec 13, 2005||Canon Kabushiki Kaisha||System, a server for a system and a machine for use in a system|
|US6985865 *||Sep 26, 2001||Jan 10, 2006||Sprint Spectrum L.P.||Method and system for enhanced response to voice commands in a voice command platform|
|US7020609 *||Mar 16, 2001||Mar 28, 2006||Texas Instruments Incorporated||Voice activated apparatus for accessing information on the World Wide Web|
|US7113981||Dec 29, 2003||Sep 26, 2006||Mixxer, Inc.||Cellular telephone download locker|
|US7117159 *||Sep 26, 2001||Oct 3, 2006||Sprint Spectrum L.P.||Method and system for dynamic control over modes of operation of voice-processing in a voice command platform|
|US7188066 *||Feb 4, 2002||Mar 6, 2007||Microsoft Corporation||Speech controls for use with a speech system|
|US7203645 *||Apr 27, 2001||Apr 10, 2007||Intel Corporation||Speech recognition system loading different recognition engines for different applications|
|US7324947 *||Sep 30, 2002||Jan 29, 2008||Promptu Systems Corporation||Global speech user interface|
|US20010019604||May 9, 2001||Sep 6, 2001||In Touch Technologies Limited, British Virgin Islands||Enhanced communication platform and related communication method using the platform|
|US20020015480||Jun 1, 2001||Feb 7, 2002||Neil Daswani||Flexible multi-network voice/data aggregation system architecture|
|US20020032564 *||Apr 19, 2001||Mar 14, 2002||Farzad Ehsani||Phrase-based dialogue modeling with particular application to creating a recognition grammar for a voice-controlled user interface|
|US20020049535||Aug 10, 2001||Apr 25, 2002||Ralf Rigo||Wireless interactive voice-actuated mobile telematics system|
|US20020106065||Apr 3, 2002||Aug 8, 2002||Joyce Simon James||Enhanced communication platform and related communication method using the platform|
|US20020146015||Mar 6, 2001||Oct 10, 2002||Bryan Edward Lee||Methods, systems, and computer program products for generating and providing access to end-user-definable voice portals|
|US20030004728 *||Jun 27, 2001||Jan 2, 2003||Keiller Robert Alexander||System|
|US20030028380||Aug 2, 2002||Feb 6, 2003||Freeland Warwick Peter||Speech system|
|US20030033152||May 30, 2002||Feb 13, 2003||Cameron Seth A.||Language independent and voice operated information management system|
|US20030065427||Mar 1, 2002||Apr 3, 2003||Karsten Funk||Method and device for interfacing a driver information system using a voice portal server|
|US20030068154||Nov 4, 2002||Apr 10, 2003||Edward Zylka||Gateway content storage system having database indexing, and method thereof|
|US20030073434||Aug 30, 2002||Apr 17, 2003||Shostak Robert E.||Voice-controlled wireless communications system and method|
|US20030093281 *||May 21, 1999||May 15, 2003||Michael Geilhufe||Method and apparatus for machine to machine communication using speech|
|US20030177013 *||Feb 4, 2002||Sep 18, 2003||Falcon Stephen Russell||Speech controls for use with a speech system|
|US20040077334||Oct 15, 2003||Apr 22, 2004||Upaid Systems||Enhanced communication platform and related communication method using the platform|
|US20040110472||Feb 21, 2003||Jun 10, 2004||Johnson Controls Technology Company||Wireless communication system and method|
|US20040127241||Dec 12, 2003||Jul 1, 2004||Vocera Communications, Inc.||Voice-controlled wireless communications system and method|
|US20040132433||Jul 16, 2001||Jul 8, 2004||Stern Robert A.||System and method for directory services and e-commerce across multi-provider networks|
|US20050143139||Dec 30, 2004||Jun 30, 2005||Samsung Electronics Co., Ltd.||Method for controlling mobile phone to output audio signals and alert sounds through external audio player|
|US20050144251||Dec 29, 2003||Jun 30, 2005||Slate Michael L.||Cellular telephone download locker|
|US20050170863||Mar 1, 2005||Aug 4, 2005||Shostak Robert E.||Voice-controlled wireless communications system and method|
|US20060018440||Jul 26, 2004||Jan 26, 2006||Watkins Gary A||Method and system for predictive interactive voice recognition|
|US20060050686||Sep 6, 2005||Mar 9, 2006||Commoca, Inc.||Software platform for developing, delivering and managing data-voice applications operating on an internet protocol (IP) phone|
|US20060085521||Sep 24, 2004||Apr 20, 2006||Sztybel Robert S||Interactive audio content delivery system and method|
|US20060206339||May 18, 2005||Sep 14, 2006||Silvera Marja M||System and method for voice-enabled media content selection on mobile devices|
|US20060206340||Feb 21, 2006||Sep 14, 2006||Silvera Marja M||Methods for synchronous and asynchronous voice-enabled content selection and content synchronization for a mobile or fixed multimedia station|
|EP1003018A2||Jul 15, 1999||May 24, 2000||Fujitsu Limited||Characteristic extraction apparatus for moving object and method thereof|
|EP1341363A1||Nov 20, 2002||Sep 3, 2003||Robert Bosch Gmbh||Method and device for interfacing a driver information system using a voice portal server|
|EP1633150A2||Sep 14, 1999||Mar 8, 2006||Upaid Systems, Ltd.||Communication services|
|EP1633151A2||Sep 14, 1999||Mar 8, 2006||Upaid Systems, Ltd.||Communication services|
|EP1742437A1||Jul 6, 2005||Jan 10, 2007||Alcatel||Provision of a telecommunication connection|
|WO2000016568A1||Sep 14, 1999||Mar 23, 2000||In Touch Technologies Limited||Communication services|
|WO2000021232A2||Oct 1, 1999||Apr 13, 2000||International Business Machines Corporation||Conversational browser and conversational systems|
|WO2001022112A1||Sep 21, 2000||Mar 29, 2001||Motorola, Inc.||A method and apparatus for providing navigational services in a wireless communication device|
|WO2001022249A1||Sep 21, 2000||Mar 29, 2001||Motorola Inc.||A method and apparatus for saving enhanced information contained in content sent to a wireless communication device|
|WO2001022633A1||Sep 21, 2000||Mar 29, 2001||Motorola Inc.||A method and apparatus for enabling the selection of content on a wireless communication device|
|WO2001022712A1||Sep 21, 2000||Mar 29, 2001||Motorola Inc.||A method and apparatus for remotely configuring a wireless communication device|
|WO2001022713A1||Sep 21, 2000||Mar 29, 2001||Motorola Inc.||A method and apparatus for remotely configuring a wireless communication device|
|WO2001039178A1||Nov 10, 2000||May 31, 2001||Koninklijke Philips Electronics N.V.||Referencing web pages by categories for voice navigation|
|WO2001057851A1||Feb 2, 2001||Aug 9, 2001||Famoice Technology Pty Ltd||Speech system|
|WO2002007050A2||Jul 16, 2001||Jan 24, 2002||Context Connect, Inc.||A system and method for directory services and e-commerce across multi-provider networks|
|WO2002011120A1||Aug 2, 2001||Feb 7, 2002||Speaklink, Inc.||System and method for voice-activated web content navigation|
|WO2002017090A1||Aug 23, 2001||Feb 28, 2002||Friedman, Mark, M.||A method and system for automatically connecting real-world entities directly to corresponding network-based data sources or services|
|WO2002097590A2||May 30, 2002||Dec 5, 2002||Cameronsound, Inc.||Language independent and voice operated information management system|
|WO2004021149A2||Aug 29, 2003||Mar 11, 2004||Vocera Communications, Inc.||Voice-controlled wireless communications system and method|
|WO2004077721A2||Feb 17, 2004||Sep 10, 2004||Johnson Controls Technology Company||Wireless communication system and method|
|WO2005079254A2||Feb 10, 2005||Sep 1, 2005||Fujitsu Limited||Wireless wallet|
|WO2006029269A2||Sep 8, 2005||Mar 16, 2006||Commoca, Inc.||Software platform for data-voice applications operating on an internet protocol (ip) phone|
|WO2006033841A2||Sep 7, 2005||Mar 30, 2006||Sztybel Robert S||Interactive audio content delivery system and method|
|WO2006098789A2||Dec 19, 2005||Sep 21, 2006||Apptera, Inc.||System and method for voice-enabled media content selection on mobile devices|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7599838 *||Sep 1, 2004||Oct 6, 2009||Sap Aktiengesellschaft||Speech animation with behavioral contexts for application scenarios|
|US8626511 *||Jan 22, 2010||Jan 7, 2014||Google Inc.||Multi-dimensional disambiguation of voice commands|
|US8635075 *||Oct 12, 2009||Jan 21, 2014||Microsoft Corporation||Raising the visibility of a voice-activated user interface|
|US8650036 *||Oct 10, 2012||Feb 11, 2014||Samsung Electronics Co., Ltd.||Electronic apparatus and method of controlling electronic apparatus|
|US8793127||Oct 31, 2007||Jul 29, 2014||Promptu Systems Corporation||Method and apparatus for automatically determining speaker characteristics for speech-directed advertising or other enhancement of speech-controlled devices or services|
|US8862596||Nov 2, 2012||Oct 14, 2014||Promptu Systems Corporation||Method and apparatus for generation and augmentation of search terms from external and internal sources|
|US8959019 *||Oct 31, 2007||Feb 17, 2015||Promptu Systems Corporation||Efficient empirical determination, computation, and use of acoustic confusability measures|
|US9106945||Apr 8, 2014||Aug 11, 2015||Samsung Electronics Co., Ltd.||Electronic apparatus and method of controlling electronic apparatus|
|US9148688||Apr 8, 2014||Sep 29, 2015||Samsung Electronics Co., Ltd.||Electronic apparatus and method of controlling electronic apparatus|
|US9305549||Oct 13, 2014||Apr 5, 2016||Promptu Systems Corporation||Method and apparatus for generation and augmentation of search terms from external and internal sources|
|US20060047520 *||Sep 1, 2004||Mar 2, 2006||Li Gong||Behavioral contexts|
|US20070028262 *||Jul 29, 2005||Feb 1, 2007||Zermatt Systems, Inc.||Virtual discovery of content available to a device|
|US20070150273 *||Dec 22, 2006||Jun 28, 2007||Hiroki Yamamoto||Information retrieval apparatus and method|
|US20080103761 *||Oct 31, 2007||May 1, 2008||Harry Printz||Method and Apparatus for Automatically Determining Speaker Characteristics for Speech-Directed Advertising or Other Enhancement of Speech-Controlled Devices or Services|
|US20080126089 *||Oct 31, 2007||May 29, 2008||Harry Printz||Efficient Empirical Determination, Computation, and Use of Acoustic Confusability Measures|
|US20080133238 *||Nov 30, 2007||Jun 5, 2008||Canon Kabushiki Kaisha||Information processing apparatus and information processing method|
|US20100031150 *||Oct 12, 2009||Feb 4, 2010||Microsoft Corporation||Raising the visibility of a voice-activated user interface|
|US20100050270 *||Feb 25, 2010||AT&T InteIlectual Property I, L.P.||Control of Access to Content Received from a Multimedia Content Distribution Network|
|US20110040563 *||Feb 5, 2010||Feb 17, 2011||Xie-Ren Hsu||Voice Control Device and Voice Control Method and Display Device|
|US20110184730 *||Jan 22, 2010||Jul 28, 2011||Google Inc.||Multi-dimensional disambiguation of voice commands|
|US20120023102 *||Jan 26, 2012||Veveo, Inc.||Methods and systems for dynamically rearranging search results into hierarchically organized concept clusters|
|DE102010054217A1||Dec 11, 2010||Jun 14, 2012||Volkswagen Aktiengesellschaft||Method for providing speech control system in vehicle, involves assigning selected clusters to function category, and producing control signal for executing function in vehicle depending on determined cluster function category pairs|
|U.S. Classification||704/255, 704/270, 704/275|
|International Classification||G10L17/00, G10L15/22, G10L15/00, H03M7/30, G10L15/26, G06Q30/00|
|Cooperative Classification||G10L17/00, H03M7/3082, G06Q30/02, G10L25/24, G10L15/26|
|European Classification||G06Q30/02, H03M7/30V, G10L15/26A, G10L17/00U|
|Oct 30, 2003||AS||Assignment|
Owner name: AGILE TV CORPORATION, CALIFORNIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:MADDUX, SCOTT LYNN;JORDAN, ADAM;WILLIAMS, JODY;AND OTHERS;REEL/FRAME:014663/0973;SIGNING DATES FROM 20031028 TO 20031029
|Jun 18, 2010||AS||Assignment|
Owner name: PROMPTU SYSTEMS CORPORATION,CALIFORNIA
Free format text: CHANGE OF NAME;ASSIGNOR:AGILE TV CORPORATION;REEL/FRAME:024555/0927
Effective date: 20060925
|Sep 19, 2012||FPAY||Fee payment|
Year of fee payment: 4